JP2016532927A - System and method for managing logical volumes - Google Patents

Abstract

The management system manages a plurality of logical volumes in the storage system connected to the host system. The plurality of logical volumes includes a first logical volume and a second logical volume provided to the host system. The management system manages the second logical volume as a replica volume of the first logical volume and as a volume in the first state that is a logical volume provided to the host system. The management system changes the second logical volume to a volume in the second state associated with the first logical volume. The volume in the second state is not a logical volume provided to the host system, but is in a standby state for future use by the host system. When data is newly stored in the first logical volume, the management system stores the data stored in the second logical volume that is the volume in the second state and the second logical volume associated with the second logical volume. A data difference from data stored in one logical volume is managed. [Selection] Figure 1

Description

  The present invention generally relates to management of logical volumes.

  In recent years, data centers have dynamically allocated virtualized resources to data center computers, networks, and storage in order to efficiently use resources and quickly eliminate or reduce bottlenecks and failures. It has become. Specifically, volume scale-out is performed to meet an expected or unexpected increase in service read workload demand that cannot be met with a single logical volume. In addition, volume migration is performed to meet the expected or unexpected increase in disk performance demand (eg, requiring a change from SATA to SSD) and balance some storage workload. The Migration may also be required to resume service when a problem occurs in the I / O path between the host and its provided logical volume. However, performing volume scale-out and volume migration takes a long time because a complete data copy is required.

  On the other hand, there is a data reuse method to avoid copying a large amount of data between logical volumes. Patent Document 1 discloses a system in which stored application data is not deleted, but is stored for future reuse by the same application, even when a command for deleting application data is used. ing. This method avoids the time consuming task of creating a backup and restoring application data that needs to be reused.

US Patent Application Publication No. 2012/0179823

  The related art disclosed in Patent Document 1 is used for the purpose of reusing data as a copy of previously discarded data. However, this technique cannot be used for logical volume scale-out or migration. This is because in these cases, at least one logical volume is still provided to the host. Therefore, in order to reuse the data once copied with respect to the original data, it is necessary to manage the relationship between the original data. For example, when a logical volume is migrated to a new logical volume, the original volume is usually deleted. Even if the data in the source logical volume is retained without being deleted, it cannot be used as a copy of the data in the destination logical volume provided to the host. This is because, in the system of Patent Document 1, (1) retained data is not associated with the original data, but is associated with the application, and (2) retained data is related to the original data. It is because it is not updated according to the change of. As a result, when performing management operations such as at least one of logical volume scale-out and logical volume migration, it is retained to avoid the task of copying logical volumes, which consumes time and resources. It is impossible to reuse data. Logical volume scale-out and logical volume migration are operations that need to be done in a short time to resolve unexpected problems and demands before downtime and poor performance occur.

  The management system manages a plurality of logical volumes in the storage system connected to the host system. The plurality of logical volumes includes a first logical volume and a second logical volume provided to the host system. The management system manages the second logical volume as a replica volume of the first logical volume and as a volume in the first state that is a logical volume provided to the host system. The management system changes the second logical volume to a volume in the second state associated with the first logical volume. The volume in the second state is not a logical volume provided to the host system, but is in a standby state for future use by the host system. When data is newly stored in the first logical volume, the management system stores the data stored in the second logical volume that is the volume in the second state and the second logical volume associated with the second logical volume. A data difference from data stored in one logical volume is managed.

  In accordance with the present invention, in a data center, an automated process or administrator performs unplanned management operations such as at least one of scale-out and migration without requiring time-consuming logical volume copies. can do.

FIG. 3 is a diagram showing the association of a host and a logical volume and the development of a logical volume state over time in a storage system in response to an executed management operation according to an embodiment of the present invention. It is a figure which shows schematic structure of the computer system which concerns on embodiment. It is a figure which shows the structural example of the logical volume correlation table provided in a management computer. It is a figure which shows the structural example of the logical volume pairing table provided in a management computer. It is a figure which shows the structural example of the logical volume management table provided in a management computer. It is a figure which shows the structural example of the storage management table provided in a management computer. FIG. 6 is a diagram illustrating a flowchart for explaining processing executed by a management computer at the time when a logical volume replication request arrives. It is a figure for demonstrating the process performed by the management computer at the time of a logical volume deletion request reaching. It is a figure which shows the flowchart for demonstrating the process periodically performed by a management computer, for example with a short time interval.

  One embodiment of the present invention is described below.

  In the following description, information is described using the expression “aaa table”, but the information can also be expressed by a data structure other than a table. To represent independence from the data structure, the term “aaa table” can be replaced with the term “aaa information”.

  An ID (identifier) is used as information for identifying a target in the following description, but the ID can be replaced with other types of identification information.

  In the following description, the term “program” may be used as a principal for describing a process. However, the program is executed by a processor unit (eg, CPU (Central Processing Unit)) that executes a process determined using at least one of a storage unit (eg, memory) and an interface unit (eg, communication port). Executed. Therefore, the process subject can be a processor unit. Part of the process can be performed by a hardware circuit, and the processor unit can comprise a hardware circuit. The program can be installed from the program source. The program source can be, for example, a program distributed server or a portable storage medium.

  In the following description, a collection of one or more computers configured to manage logical volumes and display information for display may be referred to as a “management system”. In the example in which the management computer displays information for display, the management computer can serve as a management system. Furthermore, a combination of a management computer and a display computer can also serve as a management system. To improve the speed and reliability of the management process, multiple computers can be used to achieve a process that is the same as or similar to the process performed by the management computer. In this case, a plurality of computers (may include a display computer when a display computer is used for display) can serve as a management system. In the embodiment, the management computer can serve as a management system. The phrase “the management computer displays information” can indicate that information is displayed on a display device held by the management computer, or the display information is displayed on a display computer connected to the management computer. It can also represent being transmitted. In the latter case, the display computer displays information, which is represented by information for display on a display device held by this display computer.

  In the following description, the term “logical volume” represents a logical storage device. The logical volume can be a real logical volume or a virtual logical volume. The term “real logical volume” refers to a logical volume based on one or more PDEVs. The term “PDEV” is an abbreviation for a non-volatile physical storage device such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). Multiple PDEVs can form multiple RAID (redundant arrays of independent (or inexpensive) disks) groups. A RAID group can be referred to as a “parity group”. An example of a “virtual logical volume” can include a logical volume that follows storage virtualization technology, which includes a virtual logical volume, a logical volume that follows capacity virtualization technology (usually thin provisioning), and the original logical volume. Based on a storage resource (for example, a logical volume) of an external storage device connected to a storage device having a logical volume provided as a snapshot.

<Outline of Embodiment>

  An overview of the embodiment will be described with reference to the diagram shown in FIG.

  FIG. 1 shows the state of the logical volume 2212 in the storage system at four different times. The time points are (t), (t + 1), (t + 2), and (t + 3) in order from the earliest to the later. Each time point occurs after an arbitrary period of time has elapsed since the previous time point. The logical volume has one of three states managed by the management computer (FIG. 5). The three states are “provided”, “standby” and “formatted”. The formatted logical volume is a logical volume that has no data and is not provided to a host system such as the host computer 2000 connected to the storage system. In other words, the formatted logical volume is not managed by the host computer and cannot be recognized by the host computer. (The host system comprises two host computers named Host1 and Host2 in the example according to FIG. 1). The formatted logical volume can be a formatted logical volume. The standby logical volume is a logical volume that has data and is not provided to the host system. The standby logical volume is in a standby state for future use by the host system. The provided logical volume is a logical volume that has data and is provided to the host computer 2000. That is, the host computer 2000 can have read access and / or write access to data in the provided logical volume.

  The time (t) represents a host computer named Host1 that provides the service. The provided logical volume named logical volume 1, hereinafter LV1, is connected to Host1 (FIG. 5). Two logical volumes named LV2 and LV3 remain in the formatted state.

  After time (t), the service provided by Host1 has a temporary demand for heavy read workload that cannot be met by LV1 alone. Therefore, scale out is required. Since LV1 is not associated with any standby logical volume, scale-out is executed as follows (FIG. 7). (1) In response to a request from the management system, the entire data stored in the logical volume LV1 is copied to the LV2 by the storage system. (2) After copying, autonomously by the storage system or from the management system LV2 is provided to another host computer named Host2 in response to the request (LV2 may be provided to Host1 instead of Host2). As a result, the demand is met after the logical volume is copied. This may take quite a long time depending on the size of the logical volume. Furthermore, copying LV1 to LV2 may negatively impact LV1 performance when LV1 is most needed.

  The time (t + 1) represents a state when the scale-out task is finished and LV2 is a provided logical volume connected to Host2. The heavy read workload of the service is split or balanced between Host1 and Host2. Data in LV1 and LV2 is kept the same by propagating changes (writes) reaching one of the logical volumes to the other logical volume. The change can be propagated by at least one of the storage system, the host system, and the management system. The logical volumes LV1 and LV2 are considered to be in a pairing relationship because their synchronization is maintained by the storage system. The recent read / write rate of LV1 is managed by the management system. The read / write rate is expressed as a value obtained by dividing the read frequency (for example, IOPS (I / O per second)) by the read frequency and the write frequency. Further, the storage system can manage the pairing relationship between LV1 and LV2 and keep their data synchronized.

  After time (t + 1), the service demand for heavy read workload ends. In order to use the storage system resources efficiently, Host2 is released from the service, and resources for connecting Host2 and LV2 are released. The management system determines that LV2 is suitable for becoming a standby logical volume based on the recent read / write rate of LV1. This represents that the service has a heavy read workload and no heavy write workload (FIG. 8). In addition, the management system can request the storage system to change the status of the LV2 from provided to standby. In response to the request, the storage system can manage LV2 as a standby logical volume.

  When LV2 becomes a standby logical volume, LV2 will hold the same data as LV1, but those data will be different once the write reaches LV1. The association and data difference between LV1 and LV2 are managed by at least one of the management system and the storage system (FIG. 3). When certain conditions are reached, the different data is transferred from LV1 to LV2 in order to synchronize LV1 and LV2 partially or fully by the storage system, either autonomously or in response to a request from the management system (FIG. 9). Can be copied to. When there is no more free space or formatted logical volume in the computer system, the standby logical volume LV2 is deleted or formatted (FIG. 7).

  The time point (t + 2) represents a state when LV1 and LV2 are associated and the data difference between them is managed by at least one of the management system and the storage system. LV1 is a provided logical volume, and LV2 is a standby logical volume. When the amount of data difference exceeds the threshold at time (t + 2), the managed data difference is autonomously responded to the request from the management system or by the storage system so that the amount is less than the threshold. And can be copied completely or partially from LV1 to LV2. As a result, it is possible to synchronize LV1 and LV2 after time (t + 2).

  After time (t + 2), the service provided by Host1 again has a temporary demand for heavy read workload that cannot be met by LV1 alone. Therefore, scale out is required. Since LV1 is associated with the standby logical volume, the scale-out is executed in a short time as follows (FIG. 7). (1) To completely synchronize LV1 and LV2, the managed data difference is copied from LV1 to LV2, and (2) after copying, LV2 is provided to a host named Host2. As a result, demand is met fairly quickly after copying only a small portion or any portion of the data. When LV1 is associated with several standby logical volumes, the most suitable volume to be offered is selected as the standby logical volume with the smallest data difference from LV1.

  The time point (t + 3) represents the state when the scale-out task is completed, similarly to the state represented at the time point (t + 1).

  As described in the overview above, (1) keep the logical volume requested to be deleted, (2) associate the held logical volume with its original logical volume provided to the host, (3) By managing the data difference between the logical volumes, a complete copy of the logical volume is performed on one replicated logical volume of the original logical volume when the logical volume is first replicated Can be provided multiple times. Furthermore, the management system according to the embodiment addresses the burden of determining which logical volumes are retained, how long they are retained, and when they should be reused or deleted. The management system can be used by automated processes such as autoscaling and migration, and can also be used by unskilled users and administrators who require logical volume replication and logical volume deletion operations.

  <Configuration of computer system>

  FIG. 2 is a diagram showing a schematic configuration of a computer system according to the embodiment of the present invention.

  The computer system 100 includes a storage system and a host system connected to the storage system via a network such as a SAN (Storage Area Network). A management system that manages a plurality of logical volumes 2212 is connected to the storage system and the host system via the same or different network such as a LAN (local area network).

  The host system is at least one host computer 2000. The host computer 2000 is a computer that uses the logical volume 2212 provided by the storage apparatus 2200. In particular, the host computer 2000 issues an I / O (input / output) request that designates the logical volume 2212 as the storage apparatus 2200. The host computer 200 includes an interface unit, a storage unit, an input unit, an output unit, and a processor unit connected thereto. The interface unit is at least one interface device such as a data I / F 2015 and a management I / F 2016. A data I / F 2015 (for example, a SAN interface device) is connected to the storage system. A management I / F 2016 (for example, a LAN interface device) is connected to the management system. The data I / F 2015 and the management I / F 2016 can be the same. The storage unit is at least one storage device such as the memory 2014. The input unit is at least one input device 2011 such as a keyboard and a pointing device. The output unit is at least one output device 2013 such as a display device. The processor unit is at least one processor such as the CPU 2012. The memory 2014 stores a program such as the application 2021, and the CPU 2012 executes the application 2021. The application 2021 includes a program for using the provided logical volume.

  The storage system is at least one storage device 2200. The storage apparatus 2200 includes a plurality of disk devices 2213 and a storage controller unit connected to the plurality of disk devices 2213.

  Each disk device 2213 is an example of a PDEV. The plurality of disk devices 2213 can be at least one RAID group. The plurality of logical volumes 2212 are based on a plurality of disk devices 2213.

  The storage controller unit is at least one storage controller such as the disk controller 2211. The disk controller 2211 includes an interface unit, a storage unit such as a memory 2244, and a processor unit such as a CPU 2223 connected to the interface unit and the storage unit. The interface unit includes a data I / F 2211 connected to the host system, a management I / F 2222 connected to the management system, and a disk I / F 2215 connected to a plurality of disk devices 2213. The memory 2224 stores management information 2231 and a control program 2232. The management information 2231 is information regarding a plurality of logical volumes 2212. The management information 2231 can include the same tables as the tables 2121 to 2124. The CPU 2223 executes the control program 2232. The disk controller 2211 receives an I / O request designating the logical volume 2212 and write / read data from / to the disk device 2213 that is the basis of the logical volume. The disk controller 2211 manages the data difference between the data stored in the first logical volume and the data stored in the second logical volume having a specific relationship with the first logical volume. In the embodiment, the first logical volume is a provided logical volume such as a primary logical volume (copy source logical volume) having a pairing relationship with a secondary logical volume (copy destination logical volume), or a copy pair A logical volume that is not formed but is associated with a standby logical volume. The second logical volume is a secondary logical volume having a pairing relationship with the primary logical volume, a standby logical volume associated with the primary logical volume, or a formatted logical volume. The disk controller 2211 autonomously or in response to a request from the management system executes a copy of data (all data or differential data) from one volume to another volume, thereby matching the same pair (group). ) Is configured to keep the volume in the same. Further, the disk controller 2211 autonomously or in response to a request from the management system, data stored in the provided logical volume, data stored in the standby logical volume associated with the provided logical volume, and Is configured to copy the data difference between fully or partially. Further, the disk controller 2211 is configured to input (write) and output (read) data in a logical volume designated by an I / O request from the host system and data from this logical volume. Further, the disk controller 2211 is configured to set information in the management information 2231 in response to a request from the management system. Information necessary for this type of processing is stored in the management information 2231. For example, the management information 2231 includes, for each logical volume, a volume ID, a volume type (eg, primary, secondary, provided or standby), an associated volume ID (eg, a paired volume ID or an associated standby). Information defining logical volume ID) can be included.

  As defined above, in the embodiment, the management system is the management computer 2100. The management computer 2100 includes an interface unit, a storage unit, an input unit, an output unit, and a processor unit connected thereto. The interface unit is at least one interface device such as a management I / F 2215. The management I / F 2215 is connected to the host system and the management system. The storage unit is at least one storage device such as the memory 2114. The input unit is at least one input device 2111 such as a keyboard or a pointing device. The output unit is at least one output device 2113 such as a display device. The processor unit is at least one processor such as the CPU 2112. The memory 2114 stores programs and information. The information includes (1) a logical volume association table 2121 (FIG. 3) and (2) the same data for managing the association between the provided logical volume, the standby logical volume, and their data difference. Logical volume pairing table 2122 (FIG. 4) for managing the relationship between logical volumes, (3) Logical volume management table 2123 (providing logical volume to the host, and managing the status and capacity of the logical volume) 5) and (4) a storage management table 2124 (FIG. 6) for managing the capacity of the storage device 2200. The program includes (1) a logical volume providing program 2125 (FIG. 7) for providing a logical volume to the host system at the time of request, and (2) a logical volume deletion program 2126 (FIG. 8) for coping with the logical volume deletion request. And (3) a difference update program 2127 (FIG. 9) for updating the data difference between the associated logical volumes. Information stored by the logical volume association table 2121, logical volume pairing table 2122, logical volume management table 2123, and storage management table 2124 is appropriately updated by the logical volume provision program 2125, logical volume deletion program 2126, and differential update program 2127. Referenced and updated.

  <Logical volume association table 2121>

  FIG. 3 shows a configuration example of the logical volume association table 2121 provided in the management computer 2100. The logical volume association table 2121 is a table that manages associations between the provided logical volumes, the standby logical volumes, and their data differences across all the storage apparatuses 2200. The logical volume association table 2121 includes, as configuration information, a provided logical volume ID 3001 that is information for identifying the logical volume 2212, a standby logical volume ID 3002 that is information for identifying the logical volume 2212, and two logical volumes. A data difference amount 3303 representing a data difference amount between them, a data difference 3004 which is a pointer to information of a data portion different between two logical volumes (for example, a bitmap representing a data difference), and a provided logical volume It has an important flag 3005 indicating whether it is important, and a data difference threshold 3006 indicating the maximum data difference allowed for the provided logical volume that is important. For the volume ID, for example, “V010 (VSP1)” indicates that the storage device with the storage ID “VSP1” has a logical volume with the ID “V010”.

  The row of the logical volume association table 2121 represents the association between the provided logical volume and the standby logical volume. Rows are added when the logical volume goes to standby, ie when an association is created (FIG. 8). A row is deleted when the provided logical volume and / or standby logical volume changes state or is deleted, that is, when the association is removed (FIGS. 7 and 9). The logical volume cannot be standby at the same time (FIG. 5). The provided logical volume is associated with 0, 1 or a plurality of standby logical volumes. When the provided logical volume is associated with zero standby logical volumes, the provided logical volume does not appear in the logical volume association table 2121. When a provided logical volume is associated with one or more standby logical volumes, each association to the standby logical volume appears on a different line. In the logical volume association table 2121, it can be seen that the provided logical volume having the ID V010 is associated with two standby logical volumes: V030 and V025. The standby logical volume is always associated with one provided logical volume and always appears in one row in the logical volume association table 2121.

  The standby logical volume can be associated with the provided logical volume that also appears as the primary logical volume in the logical volume pairing table 2122 (FIG. 4) in the logical volume association table 2121 by the CPU 2212. It can be seen from the logical volume association table 2121 that the logical volume having the ID “V10” is associated with the two standby logical volumes. From the logical volume pairing table 2122 (FIG. 4), the logical volume V10 is simultaneously assigned the ID “ It can be seen that this is the primary logical volume of the secondary logical volume having “V070”. However, the standby logical volume cannot be associated by the CPU 2212 with the provided logical volume that also appears as the secondary logical volume in the logical volume pairing table 2122 in the logical volume association table 2121. When the provided logical volume appearing in the logical volume association table 2121 and the logical volume pairing table 2122 changes the role from primary to secondary by the CPU 2112, the logical volume association table 2121 is also changed by the CPU 2112 accordingly, and the standby logical volume Is associated with the new primary logical volume.

  The provided logical volume ID 3001 and the standby logical volume ID 3002 identify the logical volume 2212 and the storage device 2200 to which this logical volume belongs, respectively. The logical volume 2212 has an ID assigned at the time of creation that never changes until the logical volume is deleted, that is, until it becomes a free area 6002 (FIG. 6). The logical volume ID is used in the logical volume association table 2121, the logical volume pairing table (FIG. 4), and the logical volume management table (FIG. 5).

  The data difference amount 3003 and the data difference 3004 respectively represent different amounts of data and different portions of data between the provided logical volume identified in the same row in the logical volume association table 2121 and the standby logical volume. Both values are changed by the CPU 2112 (management computer 2100) (FIG. 9). The standby logical volume has the same data as the logical volume provided at the time of association, and the standby logical volume does not change unless there is an explicit synchronization request by the CPU 2112. It can also be expressed as data that has changed in the provided logical volume from the time of synchronization or from the time of association. The data difference 3004 is a pointer to a location where information is stored, for example, the memory 2224 of the storage apparatus 2200 to which the provided logical volume belongs. Information can be stored, for example, using bitmap techniques. Several rows of the logical volume association table 2121 can point to the same information, or a combination to efficiently use memory resources.

  The important flag 3005 includes one of two values: “Yes” and “No”. A value of “Yes” indicates that the logical volume association represented by the same row in the logical volume association table 2121 is important. A value of “No” indicates that the above association is not important. This flag is created when the association is created and can be changed at any time. The flag can be automatically set by the CPU 2112 (management computer 2100), for example, in a scenario in which the management unit pre-sets the maximum performance parameter that the service should be able to reach, Determine the number of standby logical volume associations that should be set as important in order to be able to quickly meet their maximum performance.

  The data difference threshold 3006 is a value set in a row in which association is set as important by the importance flag 3005. When the data difference 3004 reaches the data difference threshold 3006 in the row having the important flag 3005 set to “Yes”, the CPU 2112 determines the data difference specified based on the data difference 3004 from the provided logical volume to the standby logical volume. Is determined to be completely or partially synchronized with the provided logical volume and the standby logical volume (FIG. 9).

  From the logical volume association table 2121, the provided logical volume having IDV010 belonging to the storage apparatus having ID “VSP1” is associated with the standby logical volume having ID “V030” belonging to the storage apparatus having ID “VSP2”, and logically There is 70 MB of different data between the volumes, the information of the different data part can be accessed by the pointer “0x00F65000”, the association is important, and the data difference threshold is set to “100 MB” I understand that.

  <Logical volume pairing table 2122>

  FIG. 4 shows a configuration example of the logical volume pairing table 2122 provided in the management computer 2100. The logical volume pairing table 2122 is a table that manages pairing between provided logical volumes and their recent read / write rates across all storage apparatuses 2200. The logical volume pairing table 2122 includes, as configuration information, a primary logical volume ID 4001 that is information for identifying the logical volume 2212, a secondary logical volume ID 4002 that is information for identifying the logical volume 2212, and read / write of the logical volume. And a logical volume read / write rate 4003 representing the rate.

  The row of the logical volume pairing table 2122 represents the pairing relationship between the primary logical volume and the secondary logical volume. The primary logical volume is copied to the secondary logical volume, and these data are stored in, for example, the storage system, host system, and Synchronization is maintained by at least one of the management systems (FIG. 7). When a logical volume is copied or its associated standby is provided, rows are added for the purpose of keeping their data synchronized (FIG. 7). The logical volume appearing in the logical volume pairing table 2122 is a provided logical volume. The provided logical volume is a primary logical volume, a secondary logical volume, or neither of them. When the provided logical volume is neither the primary logical volume nor the secondary logical volume, the provided logical volume does not appear in the logical volume pairing table 2122. The primary logical volume is paired with one or more secondary logical volumes, and each pairing is represented in one row. In the logical volume pairing table 2122, it can be seen that the primary logical volume having the ID “V001” is associated with the two standby logical volumes V003 and V012. The secondary logical volume is paired with one primary logical volume and appears in one row in the logical volume pairing table 2122.

  A row is deleted when the corresponding pairing relationship is canceled, or when the corresponding primary logical volume 4001 or secondary logical volume 4002 is deleted, or becomes standby or formatted (FIG. 8). When the primary logical volume is copied to the secondary logical volume, the primary and secondary roles are set, but these roles can be changed at any time. While the primary logical volume is paired with one or more secondary logical volumes, it is unusual to delete, put into standby, or format this primary logical volume. Typically, the pairing relationship is canceled before deletion, the secondary logical volume paired with the primary logical volume is deleted, or the role is changed.

  Similar to the logical volume ID used in the logical volume association table 2121 (FIG. 3), the primary logical volume ID 4001 and the secondary logical volume ID 4002 respectively identify the logical volume 2212 and the storage device 2200 to which this logical volume belongs.

  The logical volume read / write rate 4003 includes the recent rate of the primary logical volume identified in the same row of the logical volume pairing table 2122. The read / write rate is a value from 1 to 0 expressed as the read IOPS divided by all IOPS including the read / write IOPS. The higher the rate value, the more read IOs and fewer write IOs have a logical volume, so the logical volume is more suitable to be put on standby by the CPU 2112 (FIG. 8).

  From the logical volume pairing table 2122, the primary logical volume having ID “V010” belonging to the storage apparatus having ID “VSP1” is associated with the secondary logical volume having ID “V070” belonging to the storage apparatus having ID “VSP2”. The logical volume read / write rate is found to be “0.2”.

  <Logical volume management table 2123>

  FIG. 5 shows a configuration example of the logical volume management table 2123 provided in the management computer 2100. The logical volume management table 2123 is a table that manages the logical volumes and provision of those logical volumes to the host computer over all the storage apparatuses 2200. The logical volume management table 2123 includes, as configuration information, a logical volume ID 5001 that is information for identifying the logical volume 2212, and a capacity 5002 that is the total amount of data that can be stored in the logical volume identified based on the ID 5001. One of three values: “provided”, “standby” and “formatted”, and a status 5003 representing the status of the identified logical volume 5001 and a logical volume identified based on the ID 5001 If provided, it has a provision destination host 5004 indicating to which host computer it is provided.

  Each logical volume is represented by one row in the logical volume management table 2123. Rows are added and deleted when a logical volume is created or deleted, that is, when it becomes free space (FIG. 7). The capacity 5002 changes when the logical volume is resized.

  A logical volume having the status 5003 “formatted” is a logical volume that has no data and is not provided to the host system. A logical volume having the status 5003 “standby” is a logical volume that has data and is not provided to the host system. A logical volume having the status 5003 “provided” is a logical volume having data, and is connected to the host system. The state of the logical volume is changed by the CPU 2112 (management computer 2100) (FIGS. 7, 8, and 9).

  Similar to the logical volume ID used in the logical volume association table 2121 (FIG. 3), the logical volume ID 5001 identifies the logical volume 2212 and the storage apparatus 2200 to which this logical volume belongs.

  In addition to the configuration example of FIG. 5, the logical volume association table 2121 can manage different types of disk devices 2213 (for example, SATA, SSD), and the resources used (storage ports, switch ports, etc.) Additional information regarding provision of logical volumes to the host can be managed.

  From the logical volume management table 2123, the provided logical volume having the ID “V010” belonging to the storage apparatus having the ID “VSP1” has a capacity of “800 GB” and is provided to the host computer named “Host5”. I understand that.

  <Storage management table 2124>

  FIG. 6 shows a configuration example of the storage management table 2124 provided in the management computer 2100. The storage management table 2124 is a table for managing storage areas across all storage apparatuses 2200. The storage management table 2124 includes, as configuration information, a storage ID 6001 that is information for identifying a storage apparatus, a free area 6002 that represents the amount of free storage area that is not allocated to the logical volume in the identified storage apparatus 6001, and A free logical volume 6003 that represents the amount of free storage area that can be used for any use that is assigned to the logical volume in the identified storage apparatus 6001 and a logical volume that is assigned to the logical volume in the identified storage apparatus 6001 and is assigned to the host system. It has a provided logical volume 6004 representing the amount of storage area provided.

  Each storage device 2200 is represented by one row in the storage management table 2124. The amount of storage area in the free area 6002, the free logical volume 6003, and the provided logical volume 6004 are deleted when a logical volume is created (free area is assigned to the logical volume), the state is changed (FIG. 5, 5003). It changes when it is made (free space).

  In addition to the configuration example of FIG. 6, the storage management table 2124 can manage various types of disk devices 2213 (eg, SATA, SSD).

  From the storage management table 2124, it can be seen that the storage having the ID “VSP1” has a free storage area of 30 TB, a formatted free storage area of 2 TB, a standby free storage area of 4 TB, and a provided storage area of 50 TB. .

  <Logical volume provision processing>

  FIG. 7 is a response to a provisioning request for a new logical volume that is a copy of a provided logical volume, eg, for auto-scaling (scaling out) or migration (copying the original logical volume to a new logical volume) It is a flowchart for demonstrating the process to perform. The request can be received from the administrator via the input device 2111. The process receives as input the provided logical volume to be copied. Processing is initiated by an automated process when a condition is reached (eg, when the autoscale function is used and the performance falls below a predetermined limit). The process can also be initiated by the end user through a provisioning service or by an administrator through the management computer 2100.

  In 7001, the logical volume providing program 2125 refers to the logical volume association table 2121 (FIG. 3) and determines whether or not the input provided logical volume appears in at least one row in 3001. Appearing means that at least one standby logical volume is associated with the input provided logical volume. If no standby logical volume is associated (No in 7001), the process proceeds to 7003. When at least one standby logical volume is associated (Yes in 7001), the process proceeds to 7006.

  In 7002, the logical volume providing program 2125 refers to the logical volume association table 2121 (FIG. 3), and selects the most suitable as the output logical volume from the standby logical volumes associated with the input provided logical volume. Select a standby logical volume. The most suitable standby logical volume is selected as the standby logical volume having the smallest data difference 3004 (FIG. 3). This is because the smallest data difference suggests the shortest synchronization time when the processing reaches 7006. The type of disk device 2213 (eg, SATA, SSD), for example, a topology to avoid the current bottleneck of the computer system 100, and other standby logic associated with the same provided logical volume in the same storage device 2200 Other parameters such as the presence or absence of a volume can be taken into account.

  In 7003, the logical volume providing program 2125 refers to the storage management table 2124 (FIG. 6) to determine whether or not a new logical volume having the input provided logical volume size can be provided. . The new logical volume reaches the required size by the free capacity 6002 (FIG. 6) from one of the storage devices 2200, the formatted free logical volume 6003 (FIG. 6), or a combination of both. Can be provided if possible. If the system sets a policy to preferentially select a formatted volume over free space 6002, the system can save time for formatting a new volume. When a new logical volume can be provided (Yes in 7003), the process proceeds to 7005. When a new logical volume cannot be provided (Yes in 7003), the process proceeds to 7004.

  At 7004, the logical volume providing program 2125 refers to the logical volume association table 2121 (FIG. 3) and selects the standby logical volume that is most suitable for being deleted, that is, being freed up or formatted. . As explained above, the standby logical volume is managed to be a standby of the currently used volume, while the standby logical volume itself is not used by the host. Therefore, when there is a shortage of free space, it is possible to satisfy new resource requirements from the host using the standby volume. The criterion for selecting the most suitable standby logical volume is, for example, using the largest data difference 3004 (FIG. 3) and / or the important flag 3005 (FIG. 3). When considering that some or all of the selected logical volume is provided, consider the capacity of the computer system 100 logical volume, the type of disk 2213 (eg, SATA, SSD), topology, and current bottleneck. Can be put. When the standby logical volume is selected, it is formatted or made free space, and thus the capacity value in the storage management table 2124 (FIG. 6) changes. Next, the process returns to 7003 to determine again whether the free area or the formatted logical volume is sufficient. When there is not enough free space to provide a new logical volume having the size of the input provided logical volume and there is no standby logical volume, that is, the logical volume association table has no rows, for example, the management computer 2100 An error with an “out of area” message is thrown to the administrator of system 100 through output 2113.

  In 7005, the logical volume providing program 2125 refers to the storage management table 2124 (FIG. 6) and the logical volume management table 2123 (FIG. 5) to create a new logical volume from the free area 6002 (FIG. 6). Alternatively, a logical volume is selected as an output logical volume by selecting a formatted logical volume 6003 (FIG. 6) having the required size. In order to create a new logical volume having the required size, the formatted logical volume can be deleted, that is, freed up. When the output logical volume is selected, the entire input provided logical volume is copied to the output logical volume.

  In 7006, the logical volume provision program 2125 is selected from the input provided logical volumes with reference to the information pointed by the logical volume association table 2121 (FIG. 3) and the data difference 3004 (FIG. 3). By copying the data difference to the standby logical volume, the input provided logical volume and the output standby logical volume selected in 7002 are synchronized. A situation can arise where both logical volumes are already synchronized and therefore no copy is required. If it is determined that the synchronization takes more time than the complete copy based on the current state of the storage system and the current data difference amount 3003 of the standby logical volume, the complete copy is performed instead of the synchronization.

  In 7007, the logical volume providing program 2125 deletes the row corresponding to the output standby logical volume selected in 7002 from the logical volume association table 2121 (FIG. 3).

  In 7008, the logical volume providing program 2125 stores the input provided logical volume in the logical volume pairing table 2122 (FIG. 4) as the primary logical volume, and is selected as the logical volume output in 7002 or 7005. Add a new line representing the pair relationship that makes the volume a secondary logical volume.

  The normal situation when performing a scale-out is to keep both logical volumes in sync, and the normal situation when performing a migration is to change the role of the relationship and then in the logical volume The input logical volume is deleted before the data is changed. In an abnormal situation where the synchronization between the two logical volumes cannot be maintained, the pairing relationship is canceled.

  In a situation where the input logical volume is already a secondary logical volume paired with the primary logical volume V in the logical volume pairing table 2122 (FIG. 4), the new volume to be added is that the logical volume V is primary. It represents a logical volume, and represents a pair relationship in which the output logical volume is a secondary logical volume. This is because provided logical volumes to ensure that a pair of paired logical volumes with the same data are paired to the same primary volume and thus facilitate management of those standby logical volumes Because they cannot be primary and secondary at the same time.

  In 7009, the logical volume providing program 2125 provides the output logical volume to a new host computer (or the same host computer that has already recognized the input logical volume) as follows: (1) Logical volume In the management table 2123 (FIG. 5), the status of the output logical volume is changed to provided, (2) the storage area in the storage management table 2124 (FIG. 6) is updated, and (3) the I / O path is changed. Configure and ensure that the new host reaches the logical volume (eg, assign ports and configure permissions) to the corresponding disk controller 2211 and other resources of the storage device 2200, and A request is transmitted to the new host computer 200.

  A synchronization function corresponding to the pairing relationship is requested from the disk controller 2211 in the storage apparatus 2200 to which the primary logical volume or secondary logical volume belongs. The synchronization function may be configured in other resources.

  Further, the subprocess created by 7003 and 7004 is used when requesting provision of a logical volume that is not a copy of another logical volume. In this case, it is determined in step 7003 whether or not the required area amount is free.

  Further, when the computer system 100 or each storage device 2200 holds a predetermined percentage of formatted logical volumes from the total storage capacity, 7003 and 7004 sub-processes are used. In this case, it is determined in 7003 whether or not a predetermined percentage of formatted logical volumes exist, and if not, 7004 is executed. This process allows the system to avoid waiting for the formatting process.

  <Logical volume deletion processing>

  FIG. 8 illustrates a process for responding to a request to delete a provided logical volume having a pairing relationship for the purpose of, for example, auto scaling (reducing resources) or migration (deleting the original logical volume after copying). It is a flowchart for. The request can be received from the administrator via the input device 2111. The process receives as input the logical volume to be deleted. Processing is initiated by an automated process (e.g., autoscale) when a condition is reached. The process can also be initiated by the end user through a provisioning service or by an administrator through a management computer.

  In 8001, the logical volume deletion program 2126 refers to the logical volume pairing table 2122 (FIG. 4) to determine whether or not the input logical volume is suitable for becoming a standby logical volume. This is because if all deleted logical volumes become standby, the storage apparatus 2200 may run out of free space and formatted logical volumes, and trigger the deletion of the standby logical volume to acquire free space. Because there is. As a result, standby logical volumes that may be useful in the future are deleted and cannot be used in the future. To avoid this, only the paired logical volumes that are to be deleted are expected to be useful.

  A paired logical volume is suitable for becoming a standby when the corresponding logical volume read / write rate 4003 (FIG. 4) exceeds a predetermined threshold. The higher the rate value, the more read I / Os and fewer write I / Os have logical volumes and fewer data changes in the logical volumes. Therefore, the logical volume is more suitable for becoming a standby. The threshold value can be defined in advance by the management unit for each storage device 2200 or for all storage devices. Appropriateness can also be determined based on other parameters such as the recent data change rate of the logical volume that can be inferred from the logical volume cache, the number of previous scale-out operations, and the current amount of free space.

  When the input logical volume is appropriate (Yes in 8001), the process proceeds to 8002. When the input logical volume is not appropriate (No in 8001), the process proceeds to 8003.

  In 8002, the logical volume deletion program 2126 releases resources and updates the management computer table as follows: (1) The host computer to which the input logical volume is provided, and between the host computer and the logical volume A request to release resources (for example, storage ports and switches) that can be connected to each other is transmitted. (2) In the logical volume management table 2123 (FIG. 5), the state of the input logical volume is set to standby. Change, (3) update the storage management table 2124 (FIG. 6), (4) remove the row in the logical volume pairing table 2122 (FIG. 4) where the input logical volume appears, and (5) input Data that differs between the created logical volume and the provided logical volume. A request to trace (the progress of different data can be stored in another location) is sent to the disk controller 2211. (6) The input logical volume 3002 is paired with the input logical volume. A new row having an association with the provided logical volume 3001 that was the primary logical volume in the ring relationship and a pointer 3004 to a location for storing the progress of data for which the disk controller is different is associated with the logical volume. Created in the table 2121 (FIG. 3).

  In 8003, the logical volume deletion program 2126 requests the storage apparatus to delete the input logical volume, that is, make it a free area, or format it, that is, make it formatted. The processing updates the logical volume management table 2123 (FIG. 5) and the storage management table 2124 (FIG. 6).

  <Difference update process>

  FIG. 9 is a flowchart for explaining processing for keeping the association between the standby logical volume and the provided logical volume up-to-date. Processing is performed in the background at short time intervals. Processing can also be performed when a data change event is received. The processing can be performed by a single process or by a plurality of independent processes for each storage device 2200 or each management computer 2100.

  In 9001, the differential update program 2127 searches all standby logical volumes with reference to the logical volume association table 2121 (FIG. 3). For example, when processing is performed for each storage device 2200, the standby logical volume of the storage device 2200 is searched.

  In 9002, the differential update program 2127 repeats each of the standby logical volumes searched in 9001. Hereinafter, the standby logical volume selected for each iteration is referred to as a repeated logical volume.

  In 9003, the differential update program 2127 refers to the logical volume association table 2121 (FIG. 3), searches for the logical volume associated with the repetitive logical volume in 9002, and refers to the logical volume management table 2123 (FIG. 5). Then, it is determined whether or not the retrieved associated logical volume has been provided. When the associated logical volume has been provided (Yes in 9003), the process proceeds to 9005. If the associated logical volume has not been provided (No in 9003), that is, if it has been formatted, is in a standby state, or does not appear in the logical volume management table 2123, the process proceeds to 9004. By referring to the logical volume management table 2123 (FIG. 5), the capacities of the associated logical volume and repetitive logical volume can be compared. For example, an option when the capacity becomes different due to volume scale-up or scale-down is that the process proceeds to 9004 and the repetitive logical volume is deleted. An alternative is to resize the iterative logical volume to match the capacity and then proceed to 9005 to update the data difference.

  In 9004, the differential update program 2127 instructs the storage apparatus to delete, that is, make a repetitive logical volume, that is, a free area, or format, that is, to have formatted, in 9002. The processing updates the logical volume management table 2123 (FIG. 5) and the storage management table 2124 (FIG. 6).

  In 9005, the differential update program 2127 refers to the logical volume association table 2121 (FIG. 3), searches for the pointer 3004 to the data difference between the repeated logical volume and the associated logical volume, The data difference amount is calculated, and the corresponding data difference amount 3003 in the logical volume association table 2121 is updated.

  In 9006, the differential update program 2127 refers to the logical volume association table 2121 (FIG. 3) and determines in 9002 whether the repeated logical volume needs to be synchronized. In this embodiment, the logical volume needs to be synchronized when its importance flag 3005 is “YES” and the data difference amount 3003 reaches or exceeds the data difference threshold 3006. is there. Alternatively, all logical volumes can be the subject of a synchronized process without managing their importance. When the logical volume needs to be synchronized (Yes in 9006), the process proceeds to 9007. When the logical volume does not need to be synchronized (No at 9006), the process continues at 9002 to check the next standby logical volume.

  In 9007, the difference update program 2127 refers to the information pointed to by the data difference value in 3004 in the logical volume association table 2121 (FIG. 3) in order to synchronize at least exceeding the data difference threshold 3006. As one example, how much an iterative logical volume is synchronized with its associated logical volume depends on the workload of the computer system 100. This means that if the workload is higher than a predetermined value, the logical volume is only partially synchronized. When the workload is lower than a predetermined value, the logical volumes are completely synchronized. This is due to the fact that a logical volume may have a set of files that change frequently and a set of files that change only occasionally. At this time, when the size of the frequently changing file is less than the data difference threshold value, it is convenient to synchronize the changing file. It may not be efficient to copy files that change frequently each time. Steps 9006 and 9007 are not important to the system, but these steps keep the data difference amount 3003 below the data difference threshold 3006. Therefore, the time for preparing the volume from the standby logical volume to the host can be kept below a desired amount of time.

  As described above, according to the embodiment, the management computer 2100 requires a large amount of time to copy the entire logical volume to another logical volume to avoid many future tasks of copying the entire logical volume. And an executed task, which is a task that consumes resources. Embodiments can be implemented as a management process that is transparent to the administrator.

  Thus, by allowing management operations such as scale-out and migration to be performed and avoiding a complete copy of the logical volume, (1) resources are used efficiently and (2) downtime and performance Before a failure occurs, you can clear unexpected problems and meet demand in a short time.

  Although one embodiment has been described above, the present invention is not limited to the embodiment described above, and can be modified in various forms without departing from the scope and spirit of the present invention. Should be understood.

100 Computer system 2000 Host computer 2100 Management computer 2200 Storage device 2212 Logical volume

Claims (13)

An interface unit connected to the storage system and a host system of the storage system;
A processor unit connected to the interface unit for managing a plurality of logical volumes;
With
The plurality of logical volumes include a first logical volume and a second logical volume provided to the host system,
Based on the information, the processor
Managing the second logical volume as a replica volume of the first logical volume and as a volume in a first state that is a logical volume provided to the host system;
The second logical volume is changed to a second state volume associated with the first logical volume, and the second state volume is not a logical volume provided to the host system, In standby for future use by the host system,
When data is newly stored in the first logical volume, the data stored in the second logical volume, which is the volume in the second state, and the data associated with the second logical volume Managing the data difference with the data stored in the first logical volume,
Management system. If the managed second-state volume includes the second logical volume associated with the first logical volume, the processor unit responds to a request to replicate the first logical volume. In response, providing the second logical volume to the host system;
The management system according to claim 1. If the managed second-state volume does not include the second logical volume associated with the first logical volume, the processor unit sends the third-state volume to the host system. Offer to,
The volume in the third state is not a logical volume provided to the host system, but a formatted logical volume.
The management system according to claim 2. If the data difference exists, the processor unit sends the data difference to the second logical volume associated with the first logical volume before providing the second logical volume to the host system. Store,
The management system according to claim 2. When there are two or more second-state volumes associated with the first logical volume, the second logical volume has the smallest data difference among the two or more second logical volumes. The volume in the second state,
The management system according to claim 2. When the first logical volume is not associated with the second logical volume, the processor unit changes the second logical volume to a logical volume in a third state;
The volume in the third state is not a logical volume provided to the host system, but a formatted logical volume.
The management system according to claim 1. If the amount of the data difference exceeds a threshold, the processor completes the data difference to the second logical volume associated with the first logical volume that is the volume in the second state. Store in or partially,
The management system according to claim 1. The processor unit provides the second logical volume to the host system in response to a request to migrate the first logical volume associated with the second logical volume;
The management system according to claim 2. The processor unit changes the second logical volume to the volume in the second state in response to deleting the duplicate volume.
The management system according to claim 1. The processor unit manages the second logical volume as a volume in the second state when a read I / O rate between I / Os of the first logical volume exceeds a predetermined threshold. Configured to determine the
The management system according to claim 1. When the read I / O rate between the I / Os of the first logical volume is equal to or less than a predetermined threshold, the processor unit manages the second logical volume as a third state volume. Decide
The volume in the third state is not a logical volume provided to the host system, but a formatted logical volume.
The management system according to claim 10. A storage system connected to the host system,
Multiple storage devices that are the basis of logical volumes contained in multiple logical volumes;
A controller unit connected to the plurality of storage devices and the host system;
With
The plurality of logical volumes include a first logical volume and a second logical volume provided to the host system,
The controller unit is
Controlling the second logical volume to be a replica volume of the first logical volume, and providing the second logical volume to the host system;
Stopping controlling the second logical volume to be a replica volume of the first logical volume and providing the second logical volume to the host system;
Associating the second logical volume with the first logical volume such that the second logical volume is in a standby state for future use by the host system;
When data is newly stored in the first logical volume by the host system, the data stored in the second logical volume, which is the second state volume, and the second logical volume are stored in the second logical volume. Managing the data difference between the data stored in the associated first logical volume;
Configured as
Storage system. A method for managing a plurality of logical volumes including a first logical volume and a second logical volume provided to the host system in a storage system connected to the host system,
Managing the second logical volume as a replica volume of the first logical volume and as a volume in a first state that is a logical volume provided to the host system;
The second logical volume is changed to a second state volume associated with the first logical volume, and the second state volume is not a logical volume provided to the host system, In standby for future use by the host system,
When data is newly stored in the first logical volume, the data stored in the second logical volume, which is the volume in the second state, and the data associated with the second logical volume Managing the data difference with the data stored in the first logical volume,
Method.